Research on evolution model of road network for self-driving truck in open-pit mine

Open-pit mining environments are complex, and the development and transportation systems dynamically evolve as mining progresses. Roads exhibit unstructured characteristics, making it essential to identify and represent road networks’ status and rapidly predict their future developments for the efficient operation of intelligent open-pit mines and the safe functioning of unmanned transportation systems. Due to the advancement of intelligent mining technology, the disadvantages of manual measurement and mapping of open-pit mine road networks, such as high costs and delays, have become increasingly evident. This study uses virtual design to extract the road network at the next time point from the perspective of road evolution, obtaining future road network information to effectively address challenges such as the difficulty in optimizing the open-pit network and subsequent planning. It introduces a node evolution prediction method based on Stacked Long Short-Term Memory by analyzing the evolutionary characteristics and patterns of transportation road networks in open-pit mines. In addition, a road network evolution model for open-pit mines is established, considering the status of road networks and the mechanisms of node and segment evolution. The specific application of this model is demonstrated with typical open-pit mining transportation systems, generating structural maps of the road network at time point t + 1. The results indicated that the root mean square error and mean absolute error of node position prediction by the model are both less than 3 m. In addition, as the dataset expands, the stability and accuracy of the model improve.